Document keywords and handle them better

docs/README.md

@@ -36,6 +36,8 @@
 
 * [Important guidelines regarding reentrancy](lang/reentrancy.md)
 
+* [List of keywords](lang/keywords.md)
+
 ## Library reference
 
 * [`stdlib` module](stdlib/stdlib.md)

docs/doc_index.md

@@ -36,6 +36,8 @@
 
 * [Important guidelines regarding reentrancy](lang/reentrancy.md)
 
+* [List of keywords](lang/keywords.md)
+
 ## Library reference
 
 * [`stdlib` module](stdlib/stdlib.md)

docs/lang/keywords.md

@@ -0,0 +1,113 @@
+[< back to index](../doc_index.md)
+
+# List of keywords
+
+## Normal keywords
+
+### Top level keywords
+
+These keywords can occur at the top level of the file:
+
+    import segment
+    array struct union enum alias
+
+### Declaration modifiers
+
+These keywords occur within variable, array and function declarations:
+
+* Variable declaration modifiers: `const volatile register static stack` 
+
+* Function declaration modifiers: `asm inline noinline interrupt kernal_interrupt macro reentrant extern`
+
+* Array declaration modifiers: `const`
+
+* Common modifiers: `segment align`
+    
+### Statement keywords
+
+These keywords occur within function bodies:
+
+    if else for while do
+    goto label
+    asm
+    return
+
+### Sub-statement keywords
+
+These keywords occur within bodies of certain other statements:
+
+* for-loop directions: `to until downto parallelto parallel until`
+
+* special return-dispatch branch: `default`
+
+* loop control flow: `break continue`
+
+## Contextual keywords
+
+These are not considered keywords, but have special meaning in certain syntactical locations:
+
+* special alignment constant: `fast`
+
+* special array function: `file`
+
+* text encoding names: `z defaultz scr scrz`; for the rest, see [the list of text encodings](./text.md)  
+`default` is also a name of a text encoding, but it's also a keyword
+
+## Reserved identifiers
+
+Most of reserved identifiers are considered keywords (i.e. you cannot override them, even locally).
+Some are not, but overriding them is strongly discouraged and for highest code clarity should also be considered keywords.
+
+The keyword status of those identifiers may change in the future.
+
+### Built-in types
+
+The following identifiers are considered keywords:
+
+    byte ubyte sbyte word long
+    pointer void bool     
+    set_carry set_zero set_overflow set_negative    
+    clear_carry clear_zero clear_overflow clear_negative    
+    int16 int24 int32 int40 int 48 int56 int64 int72 int80 int88 int96 int104 int112 int120 int128    
+    unsigned8 signed8
+
+`farword` is a deprecated alias for `int24` and it is not a keyword. It will be removed in the future.
+    
+### Built-in constants
+
+The following identifiers are considered keywords:
+
+    true false
+    
+The following identifiers are not considered keywords:
+
+    nullptr nullchar
+    
+### Built-in functions and operators
+
+The following identifiers are considered keywords:
+
+    not hi lo nonet sizeof
+    
+The following identifiers are not considered keywords:
+
+    sin cos tan call
+
+## Reserved field names
+
+It is not allowed to define a struct or a union with a field with any of the following names:
+
+    pointer addr rawaddr return
+
+## Preprocesor keywords
+
+### Directives
+
+    #if #elseif #else #endif
+    #define #use
+    #pragma
+    #fatal #error #warn #info #infoeval
+    
+### Built-in functions
+
+    not hi lo defined if

mkdocs.yml

@@ -30,6 +30,7 @@ nav:
     - Inline 6502 assembly: lang/assembly.md
     - Inline 8080/LR35902/Z80 assembly: lang/assemblyz80.md
     - Reentrancy guidelines: lang/reentrancy.md
+    - List of keywords: lang/keywords.md
   - Library reference:
     - stdlib module: stdlib/stdlib.md
     - string module: stdlib/string.md

src/main/scala/millfork/env/Environment.scala

@@ -505,8 +505,14 @@ class Environment(val parent: Option[Environment], val prefix: String, val cpuFa
   }
 
   def assertNotDefined(name: String, position: Option[Position]): Boolean = {
-    if (builtinsAdded && Environment.keywords(name)) {
-      log.error(s"Cannot redefine a builtin keyword `$name`", position)
+    if (builtinsAdded && Environment.invalidNewIdentifiers(name)) {
+      if (Environment.predefinedFunctions(name)) {
+        log.error(s"Cannot redefine a builtin predefined function `$name`", position)
+      } else if (Environment.neverIdentifiers(name)) {
+        log.error(s"Cannot the keyword `$name` as a name", position)
+      } else if (Environment.invalidNewIdentifiers(name)) {
+        log.error(s"Cannot redefine a builtin predefined identifier `$name`", position)
+      }
       false
     } else if (things.contains(name) || parent.exists(_.things.contains(name)) || things.contains(name.stripPrefix(prefix))) {
       if (!name.contains('.')){
@@ -2120,13 +2126,30 @@ class Environment(val parent: Option[Environment], val prefix: String, val cpuFa
 }
 
 object Environment {
+  // built-in special-cased functions; can be considered keywords by some:
   val predefinedFunctions: Set[String] = Set("not", "hi", "lo", "nonet", "sizeof")
-  val keywords: Set[String] = Set(
-    "true", "false",
-    "byte", "sbyte", "word", "pointer", "void", "long",
-    "array", "const", "alias", "import", "static", "register", "stack", "volatile", "asm", "extern", "kernal_interrupt", "interrupt",
-    "for", "if", "do", "while", "else", "return", "default", "to", "until", "paralleluntil", "parallelto", "downto",
+  // built-in special-cased functions, not keywords, but assumed to work almost as such:
+  val specialFunctions: Set[String] = Set("sin", "cos", "tan", "call")
+  // keywords:
+  val neverIdentifiers: Set[String] = Set(
+    "array", "const", "alias", "import", "static", "register", "stack", "volatile", "asm", "extern", "kernal_interrupt", "interrupt", "reentrant", "segment",
+    "struct", "union", "enum",
+    "for", "if", "do", "while", "else", "return", "default",
+    "to", "until", "paralleluntil", "parallelto", "downto",
+    "break", "continue",
     "inline", "noinline"
   ) ++ predefinedFunctions
+  // predefined identifiers that cannot be overridden and do not name a type:
+  val neverValidTypeIdentifiers: Set[String] = Set(
+    "true", "false",
+  ) ++ neverIdentifiers
+  // predefined type identifiers:
+  val invalidNewIdentifiers: Set[String] = Set(
+    "byte", "sbyte", "word", "pointer", "void", "long", "bool",
+    "set_carry", "set_zero", "set_overflow", "set_negative",
+    "clear_carry", "clear_zero", "clear_overflow", "clear_negative",
+    "int8", "int16", "int24", "int32", "int40", "int48", "int56", "int64", "int72", "int80", "int88", "int96", "int104", "int112", "int120", "int128",
+    "signed8", "unsigned16") ++ neverValidTypeIdentifiers
+  // built-in special-cased field names; can be considered keywords by some:
   val invalidFieldNames: Set[String] = Set("addr", "rawaddr", "pointer", "return")
 }

src/main/scala/millfork/parser/MfParser.scala

@@ -486,7 +486,7 @@ abstract class MfParser[T](fileId: String, input: String, currentDirectory: Stri
     bank <- bankDeclaration
     flags <- functionFlags ~ HWS
     returnType <- identifier ~ SWS
-    if !InvalidReturnTypes(returnType)
+    if !Environment.neverValidTypeIdentifiers(returnType)
     name <- identifier ~ HWS
     params <- "(" ~/ AWS ~/ (if (flags("asm")) asmParamDefinition else paramDefinition).rep(sep = AWS ~ "," ~/ AWS) ~ AWS ~ ")" ~/ AWS
     alignment <- alignmentDeclaration(fastAlignmentForFunctions).? ~/ AWS
@@ -735,5 +735,4 @@ object MfParser {
 
   val functionFlags: P[Set[String]] = flags_("asm", "inline", "interrupt", "macro", "noinline", "reentrant", "kernal_interrupt")
 
-  val InvalidReturnTypes: Set[String] = Set("enum", "alias", "array", "const", "stack", "register", "static", "volatile", "import", "struct", "union")
 }